yeptris 0.1.14.2 → 0.1.15.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c3f917a019fa1768553b841e4fcbcc72d736b9e47c81980eb5a2bcd029d05790
-  data.tar.gz: 56e41fcf4c12527877d50d7178d830341656975ff57c5b6d5292c1ff7c77088a
+  metadata.gz: d146f3183f11d9eb980be238bec31b1f7c801c5a72480318a3cbb625d81ca8a1
+  data.tar.gz: 69416e379b25969d66e7f331db7a8da17dd43fd1cff315b802df9f7babd8adb3
 SHA512:
-  metadata.gz: 6d85cb48948548f7d22ad9f9339a0c7634bfe9d82bd2806c1a3579047bf59de74a64a18204977af12fa98cd8315fa39f0bc11506f571d64b9fd0f2ed701612af
-  data.tar.gz: e46055db7592ebe92447134c377a0d3bb5e862e1a4b71ccaec918fa1335db1d36c87f76a2870827e35ee9c935fe0c512e61a03696e45112104cfd573c1ec544a
+  metadata.gz: 941936e4141c052a1c5faddd116c746333e35e1de32d6ac326dde1db667986a2fbcd658eb3665cecdb66568331f5e66c12ebc7e0443c5e11958129d03f48c8fd
+  data.tar.gz: 6d62eeaf6c3c1277ee44d9b54f5a3346d7f12ed4cdaa0665ae5641beff714d8ce3070d56068217a4626d3a4cd01a128c0cea77ce90b99377b736c9807b513833

data/ext/yeptris_native/json_ruby.c

@@ -26,6 +26,8 @@ typedef struct {
     int depth;
     int err;
     rb_encoding* enc;
+    int strict_dup; /* json gem >= 3: duplicate keys raise (issue #37) */
+    VALUE dup_key;
     kcent kc[YEP_KC];
 } jr;
 
@@ -175,8 +177,14 @@ static VALUE jr_object_body(jr* j, VALUE h_pre) {
         j->i++;
         VALUE val = jr_value(j);
         if (j->err) goto out;
-        if (yep_ins_mode == YEP_INS_ASET) {
-            rb_hash_aset(h_pre == Qundef ? (h_pre = rb_hash_new_capa(8)) : h_pre, key, val);
+        if (yep_ins_mode == YEP_INS_ASET || j->strict_dup) {
+            VALUE h = h_pre == Qundef ? (h_pre = rb_hash_new_capa(8)) : h_pre;
+            if (j->strict_dup && !NIL_P(rb_hash_aref(h, key))) {
+                j->err = -3;
+                j->dup_key = key;
+                goto out;
+            }
+            rb_hash_aset(h, key, val);
         } else {
             if (pn + 2 > pcap) {
                 size_t ncap = pcap * 2;
@@ -312,10 +320,11 @@ double yep_rb_scan_time(const char* p, size_t len, int n) {
     return (double)(t1.tv_sec - t0.tv_sec) + (double)(t1.tv_nsec - t0.tv_nsec) / 1e9;
 }
 
-VALUE yep_rb_parse_json(const char* p, size_t len) {
+VALUE yep_rb_parse_json(const char* p, size_t len, int strict_dup) {
     jr j;
     memset(&j, 0, sizeof(j));
     j.p = p; j.len = len; j.enc = rb_utf8_encoding();
+    j.strict_dup = strict_dup;
     VALUE already = Qtrue;
     if (yep_gc_mode != YEP_GC_NONE) already = rb_gc_disable();
     VALUE v = jr_value(&j);
@@ -325,6 +334,10 @@ VALUE yep_rb_parse_json(const char* p, size_t len) {
     if (already == Qfalse) rb_gc_enable();
     if (yep_gc_mode == YEP_GC_START && j.err == 0) rb_gc_start();
     if (j.err == -1) rb_raise(rb_eNoMemError, "yeptris native json");
+    if (j.err == -3) {
+        rb_raise(rb_path2class("Yeptris::ParseError"), "duplicate key \"%s\" in JSON object",
+                 RSTRING_PTR(j.dup_key));
+    }
     if (j.err != 0) rb_raise(rb_path2class("Yeptris::ParseError"), "native json parse failed");
     return v;
 }

data/ext/yeptris_native/yeptris_native.c

@@ -229,7 +229,7 @@ static const YeptrisVisitVTable k_vt = {
 };
 
 /* fused JSON→Ruby (json_ruby.c) — no vtable, beats JSON.parse */
-VALUE yep_rb_parse_json(const char* p, size_t len);
+VALUE yep_rb_parse_json(const char* p, size_t len, int strict_dup);
 
 static VALUE ctx_result(rb_ctx* c, YeptrisStatus st) {
     if (st != YEPTRIS_OK || c->failed) {
@@ -244,10 +244,12 @@ static VALUE ctx_result(rb_ctx* c, YeptrisStatus st) {
     return c->root;
 }
 
-static VALUE native_load_json(VALUE self, VALUE input) {
+static VALUE native_load_json(int argc, VALUE* argv, VALUE self) {
     (void)self;
+    VALUE input, strict;
+    rb_scan_args(argc, argv, "11", &input, &strict);
     StringValue(input);
-    return yep_rb_parse_json(RSTRING_PTR(input), (size_t)RSTRING_LEN(input));
+    return yep_rb_parse_json(RSTRING_PTR(input), (size_t)RSTRING_LEN(input), RTEST(strict));
 }
 
 static VALUE native_load(VALUE self, VALUE input, VALUE schema) {
@@ -412,7 +414,7 @@ RUBY_FUNC_EXPORTED void Init_native(void) {
     utf8_enc = rb_utf8_encoding();
     VALUE mYep = rb_define_module("Yeptris");
     VALUE mNat = rb_define_module_under(mYep, "Native");
-    rb_define_singleton_method(mNat, "load_json", native_load_json, 1);
+    rb_define_singleton_method(mNat, "load_json", native_load_json, -1);
     rb_define_singleton_method(mNat, "load", native_load, 2);
     rb_define_singleton_method(mNat, "load_stream", native_load_stream, 2);
     rb_define_singleton_method(mNat, "gc_mode", native_gc_mode, 0);

data/lib/yeptris/json.rb

@@ -17,6 +17,13 @@ module Yeptris
     class Error < ::Yeptris::Error; end
     class ParseError < Error; end
 
+    # json gem 3.0 made duplicate keys an error by DEFAULT; 2.x is
+    # last-wins. The parity target is the RESOLVED json gem's own
+    # behavior — strictness follows it (issue #37, found by canon's
+    # CI where json 3.0.0 resolved while the dev box had 2.x).
+    require "json"
+    STRICT_DUPLICATE_KEYS = Gem::Version.new(::JSON::VERSION) >= Gem::Version.new("3")
+
     module_function
 
     def load(source)
@@ -24,7 +31,7 @@ module Yeptris
       source = source.to_s
       if defined?(::Yeptris::Native)
         begin
-          return ::Yeptris::Native.load_json(source)
+          return ::Yeptris::Native.load_json(source, STRICT_DUPLICATE_KEYS)
         rescue ::Yeptris::ParseError => e
           raise ParseError, e.message
         end
@@ -52,7 +59,7 @@ module Yeptris
         raise ParseError, ::Yeptris::FFI.last_error_message if st != ::Yeptris::FFI::OK
 
         begin
-          walk_strict(cols)
+          walk_strict(cols, STRICT_DUPLICATE_KEYS)
         ensure
           ::Yeptris::FFI.yeptris_value_free_columns(cols)
         end
@@ -65,7 +72,7 @@ module Yeptris
     # is the strict-JSON one (no ':sym' scan, no y/n quirk, no
     # dot-required floats). Anchors/aliases/timestamps cannot occur
     # in strict JSON — reaching them is an internal error.
-    def walk_strict(cols)
+    def walk_strict(cols, strict_dup = STRICT_DUPLICATE_KEYS)
       n = cols[:count]
       kinds = cols[:kinds].read_bytes(n).unpack("C*")
       ikeys = cols[:is_keys].read_bytes(n).unpack("C*")
@@ -78,6 +85,7 @@ module Yeptris
 
       docs = []
       stack = []
+      key_sets = strict_dup ? [{}] : nil
       pending_key = nil
       i = 0
       while i < n
@@ -88,13 +96,19 @@ module Yeptris
           place(docs, stack, pending_key) { [] }
           pending_key = nil
         when ValueML::MAP_OPEN
+          key_sets&.push({})
           place(docs, stack, pending_key) { {} }
           pending_key = nil
         when ValueML::CLOSE
           stack.pop
+          key_sets&.pop
         when ValueML::V_STR
           text = arena.byteslice(offs[i], lens[i])
           if ikeys[i] == 1 && !stack.empty? && stack.last.is_a?(Hash)
+            if strict_dup && key_sets.last.key?(text)
+              raise ParseError, %(duplicate key "#{text}" in JSON object)
+            end
+            key_sets&.last&.store(text, true)
             pending_key = text
           else
             # Records carry int64 payloads: an integer-beyond-int64

data/lib/yeptris/node.rb

@@ -340,7 +340,16 @@ class Yeptris::Node
         to_f
       end
     when :timestamp then ::Yeptris::Materializer.parse_timestamp(value)
-    else value
+    else
+      # beyond int64 the C resolver leaves plain scalars :str; Psych
+      # materializes Integer (issue #31) — rebuild from the text
+      text = value.to_s
+      if style == :plain && text.length > 18 &&
+         ::Yeptris::ValueML::PSYCH_INT_SHAPE.match?(text)
+        text.delete(",_").to_i
+      else
+        value
+      end
     end
   end
 

data/lib/yeptris/psych/parser.rb

@@ -28,7 +28,7 @@ module Yeptris
         dispatch(flat, arena)
         self
       rescue ParseError => e
-        raise SyntaxError, e.message
+        raise SyntaxError.from_parse_error(e)
       end
 
       private

data/lib/yeptris/psych.rb

@@ -30,13 +30,30 @@ module Yeptris
     # object instance exists.
     autoload :Encodable, "yeptris/psych/encodable"
     class Error < StandardError; end
+    # Psych's exact interface (issue #32): same constructor arity,
+    # same reader set (file/line/column/offset/problem/context), same
+    # message shape — drop-in consumers' rescues and constructors
+    # keep working after the rebind.
     class SyntaxError < Error
-      attr_reader :line, :column
+      attr_reader :file, :line, :column, :offset, :problem, :context
 
-      def initialize(message, line = 0, column = 0)
-        super(message)
+      def initialize(file = nil, line = 0, column = 0, offset = 0, problem = nil, context = nil)
+        @file = file
         @line = line
         @column = column
+        @offset = offset
+        @problem = problem
+        @context = context
+        where = file ? "(#{file})" : "(<unknown>)"
+        detail = context ? "#{problem} #{context}" : problem.to_s
+        super("#{where}: #{detail} at line #{line} column #{column}")
+      end
+
+      # Structured lift from the C parser's message (carries
+      # "line L, column C" detail).
+      def self.from_parse_error(error)
+        md = /\bline (\d+),? column (\d+)/.match(error.message)
+        new(nil, md ? md[1].to_i : 0, md ? md[2].to_i : 0, 0, error.message)
       end
     end
     class BadAlias < Error; end
@@ -84,7 +101,7 @@ module Yeptris
 
         Nodes::Builder.document(doc)
       rescue Yeptris::ParseError => e
-        raise SyntaxError, e.message
+        raise SyntaxError.from_parse_error(e)
       end
 
       def parse_stream(yaml)
@@ -101,7 +118,7 @@ module Yeptris
         )
         stream
       rescue Yeptris::ParseError => e
-        raise SyntaxError, e.message
+        raise SyntaxError.from_parse_error(e)
       ensure
         doc&.free if doc && !stream
       end

data/lib/yeptris/valueml.rb

@@ -14,6 +14,12 @@ module Yeptris
     V_INT = 3
     V_FLOAT = 4
     V_STR = 5
+
+  # Psych's integer shape (scalar_scanner): sign optional, no
+  # leading zero, '_' / ',' separators only between digits. Beyond
+  # int64 the C resolver leaves the scalar a string; Psych
+  # materializes Integer (issue #31) — every walk rebuilds it.
+  PSYCH_INT_SHAPE = /\A[-+]?(?:0|[1-9](?:[0-9]|,[0-9]|_[0-9])*)\z/.freeze
     V_TS = 6
     SEQ_OPEN = 7
     MAP_OPEN = 8
@@ -149,7 +155,10 @@ module Yeptris
           text = arena.byteslice(flat[i + OFF], flat[i + LEN])
           # implicit-plain ':name' scans to a Symbol (Psych's
           # ScalarScanner); quoted ':x' stays a String
-          v = if flat[i + B] == 1 && text.length > 1 &&
+          v = if flat[i + B] == 1 && text.length > 18 &&
+                PSYCH_INT_SHAPE.match?(text)
+                text.delete(",_").to_i
+              elsif flat[i + B] == 1 && text.length > 1 &&
                 text.start_with?(":") && !text.start_with?("::")
                 text[1..].to_sym
               else
@@ -290,7 +299,10 @@ module Yeptris
         case kinds[i]
         when V_STR
           text = arena.byteslice(offs[i], lens[i])
-          v = if bools[i] == 1 && text.length > 1 &&
+          v = if bools[i] == 1 && text.length > 18 &&
+                PSYCH_INT_SHAPE.match?(text)
+                text.delete(",_").to_i
+              elsif bools[i] == 1 && text.length > 1 &&
                 text.start_with?(":") && !text.start_with?("::")
                 text[1..].to_sym
               else

data/lib/yeptris.rb

@@ -3,7 +3,7 @@
 module Yeptris
   # The gem's version lives in the parent namespace's file — the last
   # internal require (yeptris/version) retired with it.
-  VERSION = "0.1.14.2".freeze
+  VERSION = "0.1.15.1".freeze
   # The error hierarchy lives in THIS file (the parent namespace's
   # own file): nested constants do not trigger a parent-constant
   # autoload, and the law forbids internal requires — defining the

metadata

@@ -1,13 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: yeptris
 version: !ruby/object:Gem::Version
-  version: 0.1.14.2
+  version: 0.1.15.1
 platform: ruby
 authors:
 - Ribose Inc.
+autorequire:
 bindir: bin
 cert_chain: []
-date: 1980-01-02 00:00:00.000000000 Z
+date: 2026-09-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -57,6 +58,7 @@ metadata:
   homepage_uri: https://github.com/leptris/yeptris
   source_code_uri: https://github.com/leptris/yeptris
   changelog_uri: https://github.com/leptris/yeptris/releases
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -71,7 +73,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 4.0.16
+rubygems_version: 3.5.22
+signing_key:
 specification_version: 4
 summary: 'The YAML counterpart of libleptris: ultra-performance YAML 1.2 for Ruby'
 test_files: []